Re: [PATCH v6 6/8] arm64: dts: qcom: sc8280xp: Add multiport controller node for SC8280

[Andrew Halaney <ahalaney@redhat.com>]

On Sat, Apr 22, 2023 at 09:38:44PM +0530, Krishna Kurapati PSSNV wrote:
> 
> 
> On 4/16/2023 12:34 AM, Krishna Kurapati PSSNV wrote:
> > 
> > 
> > On 4/14/2023 9:15 PM, Andrew Halaney wrote:
> > > On Wed, Apr 05, 2023 at 06:27:57PM +0530, Krishna Kurapati wrote:
> > > > Add USB and DWC3 node for tertiary port of SC8280 along with multiport
> > > > IRQ's and phy's. This will be used as a base for SA8295P and SA8295-Ride
> > > > platforms.
> > > > 
> > > > Signed-off-by: Krishna Kurapati <quic_kriskura@quicinc.com>
> > > > ---
> > > > Link to v5: https://lore.kernel.org/all/20230310163420.7582-7-quic_kriskura@quicinc.com/
> > > > 
> > > >   arch/arm64/boot/dts/qcom/sc8280xp.dtsi | 58 ++++++++++++++++++++++++++
> > > >   1 file changed, 58 insertions(+)
> > > > 
> > > > diff --git a/arch/arm64/boot/dts/qcom/sc8280xp.dtsi
> > > > b/arch/arm64/boot/dts/qcom/sc8280xp.dtsi
> > > > index 42bfa9fa5b96..7b81f2b0449d 100644
> > > > --- a/arch/arm64/boot/dts/qcom/sc8280xp.dtsi
> > > > +++ b/arch/arm64/boot/dts/qcom/sc8280xp.dtsi
> > > > @@ -3108,6 +3108,64 @@ usb_1_role_switch: endpoint {
> > > >               };
> > > >           };
> > > > +        usb_2: usb@a4f8800 {
> > > > +            compatible = "qcom,sc8280xp-dwc3", "qcom,dwc3";
> > > > +            reg = <0 0x0a4f8800 0 0x400>;
> > > > +            #address-cells = <2>;
> > > > +            #size-cells = <2>;
> > > > +            ranges;
> > > > +
> > > > +            clocks = <&gcc GCC_CFG_NOC_USB3_MP_AXI_CLK>,
> > > > +                 <&gcc GCC_USB30_MP_MASTER_CLK>,
> > > > +                 <&gcc GCC_AGGRE_USB3_MP_AXI_CLK>,
> > > > +                 <&gcc GCC_USB30_MP_SLEEP_CLK>,
> > > > +                 <&gcc GCC_USB30_MP_MOCK_UTMI_CLK>,
> > > > +                 <&gcc GCC_AGGRE_USB_NOC_AXI_CLK>,
> > > > +                 <&gcc GCC_AGGRE_USB_NOC_NORTH_AXI_CLK>,
> > > > +                 <&gcc GCC_AGGRE_USB_NOC_SOUTH_AXI_CLK>,
> > > > +                 <&gcc GCC_SYS_NOC_USB_AXI_CLK>;
> > > > +            clock-names = "cfg_noc", "core", "iface", "sleep",
> > > > "mock_utmi",
> > > > +                      "noc_aggr", "noc_aggr_north",
> > > > "noc_aggr_south", "noc_sys";
> > > > +
> > > > +            assigned-clocks = <&gcc GCC_USB30_MP_MOCK_UTMI_CLK>,
> > > > +                      <&gcc GCC_USB30_MP_MASTER_CLK>;
> > > > +            assigned-clock-rates = <19200000>, <200000000>;
> > > > +
> > > > +            interrupts-extended = <&pdc 127 IRQ_TYPE_EDGE_RISING>,
> > > > +                        <&pdc 126 IRQ_TYPE_EDGE_RISING>,
> > > > +                        <&pdc 16 IRQ_TYPE_LEVEL_HIGH>;
> > > > +
> > > > +            interrupt-names = "dp_hs_phy_irq", "dm_hs_phy_irq",
> > > > +                        "ss_phy_irq";
> > > > +
> > > 
> > > This is breaking the current schema (with the full series applied),
> > > I am not sure if a pwr_event IRQ exists or but it maybe necessary to
> > > modify qcom,dwc3.yaml in order to explain hardware if it doesn't exist:
> > > 
> > > (dtschema) ahalaney@halaney-x13s ~/git/linux-next
> > > (git)-[718f2024524f] % make CHECK_DTBS=y
> > > DT_SCHEMA_FILES=/usb/qcom,dwc3.yaml qcom/sa8540p-ride.dtb                                                                                  
> > > :(
> > >    LINT    Documentation/devicetree/bindings
> > >    CHKDT   Documentation/devicetree/bindings/processed-schema.json
> > >    SCHEMA  Documentation/devicetree/bindings/processed-schema.json
> > >    DTC_CHK arch/arm64/boot/dts/qcom/sa8540p-ride.dtb
> > > /home/ahalaney/git/linux-next/arch/arm64/boot/dts/qcom/sa8540p-ride.dtb: usb@a4f8800: interrupt-names:0: 'pwr_event' was expected
> > >     From schema: /home/ahalaney/git/linux-next/Documentation/devicetree/bindings/usb/qcom,dwc3.yaml
> > > /home/ahalaney/git/linux-next/arch/arm64/boot/dts/qcom/sa8540p-ride.dtb: usb@a4f8800: interrupt-names:1: 'dp_hs_phy_irq' was expected
> > >     From schema: /home/ahalaney/git/linux-next/Documentation/devicetree/bindings/usb/qcom,dwc3.yaml
> > > /home/ahalaney/git/linux-next/arch/arm64/boot/dts/qcom/sa8540p-ride.dtb: usb@a4f8800: interrupt-names:2: 'dm_hs_phy_irq' was expected
> > >     From schema: /home/ahalaney/git/linux-next/Documentation/devicetree/bindings/usb/qcom,dwc3.yaml
> > > /home/ahalaney/git/linux-next/arch/arm64/boot/dts/qcom/sa8540p-ride.dtb: usb@a4f8800: interrupt-names: ['dp_hs_phy_irq', 'dm_hs_phy_irq', 'ss_phy_irq'] is too short
> > >     From schema: /home/ahalaney/git/linux-next/Documentation/devicetree/bindings/usb/qcom,dwc3.yaml
> > > /home/ahalaney/git/linux-next/arch/arm64/boot/dts/qcom/sa8540p-ride.dtb: usb@a4f8800: interrupts-extended: [[99, 127, 1], [99, 126, 1], [99, 16, 4]] is too short
> > >     From schema: /home/ahalaney/git/linux-next/Documentation/devicetree/bindings/usb/qcom,dwc3.yaml
> > > make CHECK_DTBS=y DT_SCHEMA_FILES=/usb/qcom,dwc3.yaml
> > > qcom/sa8540p-ride.dtb  22.61s user 0.54s system 99% cpu 23.172 total
> > > (dtschema) ahalaney@halaney-x13s ~/git/linux-next (git)-[718f2024524f] %
> > > 
> > > Thanks,
> > > Andrew
> > > 
> > 
> > Hi Andrew,
> > 
> >   Thanks for pointing it out. Let me check and get back on the
> > pwr_event_irq.
> > 
> > Probably I might have missed it 😅. If so, will make sure to add it in
> > next version.
> > 
> > Regards,
> > Krishna,
> 
> 
> Hi Andrew, Johan,
> 
>   I was looking at the pwr_event_irq interrupts for Multiport controller and
> see that there are two of them as per HW specs. All targets till date have
> only 1 pwr_event_irq required.
> 
> The reason I thought I missed pwr_event_irq in my patches is because in
> downstream this is a required IRQ for all targets, so I was under assumption
> that we need it for upstream targets as well. But upstream qcom driver
> doesn't have support for this IRQ yet. And this has been made a required one
> only for SC8280 [1]/[2].
> 
> Probably we can proceed in one of the following ways:
> 1. Remove pwr_event_irq in both bindings and DT as driver support is not
> present currently.
> 2. Update the bindings for SC8280 to include an optional secondary
> pwr_event_irq for multiport controller.
> 
> I would prefer option-1 as removing them would be better because they are
> not being used. Please let me know your thoughts on this.
> 
> [1]:
> https://lore.kernel.org/all/20220713131340.29401-2-johan+linaro@kernel.org/
> [2]:
> https://lore.kernel.org/all/20220713131340.29401-6-johan+linaro@kernel.org/
> 

Personally, I prefer option 2 since the IRQ does exist technically
(although it isn't currently used), I like it being described... it
makes the dt-binding a more complete description of the hardware.

I am unsure of the rules wrt dt-bindings and usage in drivers, but I
always like to view it as "this is a description of the hardware", and
the driver bit is just nice to have to ensure that whoever is adding the
binding is actually describing things sufficiently.

You could probably add a new compatible string for the multiport
sc8280xp IP as well, and make the second IRQ non-optional in dt-binding
evaluation? That seems more inline with reality, the regular IP has 1
pwr_event_irq, multiport on this platform has 2, they behave slightly
differently and thus they deserve their own string to match on despite
being on the same platform.

Please note, I'm just a contributor -- I would not be surprised to find
out that my view is not aligned with what maintainers here think, but
that's my interpretation of things!

Hope that helps,
Andrew